Long-term co-circulation of multiple influenza A viruses in pigs, Guangxi, China

ABSTRACT

Influenza A viruses (IAVs) pose a persistent potential threat to human health because of the spillover from avian and swine infections. Extensive surveillance was performed in 12 cities of Guangxi, China, during 2018 and 2023. A total of 2540 samples (including 2353 nasal swabs and 187 lung tissues) were collected from 18 pig farms with outbreaks of respiratory disease. From these, 192 IAV-positive samples and 19 genomic sequences were obtained. We found that the H1 and H3 swine influenza A viruses (swIAVs) of multiple lineages and genotypes have continued to co-circulate during that time in this region. Genomic analysis revealed the Eurasian avian-like H1N1 swIAVs (G4) still remained predominant in pig populations. Strikingly, the novel multiple H3N2 genotypes were found to have been generated through the repeated introduction of the early H3N2 North American triple reassortant viruses (TR H3N2 lineage) that emerged in USA and Canada in 1998 and 2005, respectively. Notably, when the matrix gene segment derived from the H9N2 avian influenza virus was introduced into endemic swIAVs, this produced a novel quadruple reassortant H1N2 swIAV that could pose a potential risk for zoonotic infection.

KEYWORDS:

• Swine influenza virus
• Eurasian avian-like H1N1 and human-like H3N2
• multiple reassortment
• H9N2 avian influenza virus
• pandemic potential

Introduction

Pigs are considered intermediate hosts or potential “mixing vessels” for influenza A viruses (IAVs) and this allows the reassortment of avian and mammalian IAVs to generate pandemic-type viruses with novel biological properties. Historically, several outbreaks of pandemic IAVs were caused by reassortants derived from human, avian and swine influenza viruses, including the pandemics of 1957, 1968 and 2009 [1–3]. After the 2009 pandemic H1N1 viruses (pdm09/H1N1) spread globally in humans, it was re-introduced into pig herds, which resulted in a change of the genomic landscape of IAVs with respect to both their epidemiology and ecology. The stable internal gene cassettes from the pdm09/H1N1 lineage became predominant and reassortant with other circulating swIAVs were common [4–7]. In particular, the novel reassortments between pdm09/H1N1 and the Eurasian avian-like H1N1 (EA H1N1) were frequently detected in pigs in China [8–11], which raised concerns over the ability of these pdm09-reassorted IAVs to be potentially infectious in humans.

Several lines of evidence showed that EA H1N1 viruses owned the ability to bind to human-type receptors [12] and there were at least 36 human infections with EA H1N1 recorded [11]. In particular, the EA H1N1 swIAVs (genotypes 4 and 5) carrying internal gene segments from pdm09/H1N1 and the triple-reassortant (TR) lineage have become established in the pig population in China [11], and one of these had acquired increased human infectivity [9]. In addition, the pdm/09-like internal gene cassette was successfully incorporated into the swine H3N2 virus, and it has carried the surface genes from the recent human-like H3N2 (HL H3N2) lineage since 2010 [10, 13, 14]. Importantly, the swine H3 IAVs were found to have been derived from human-like viruses, which had a higher potential to infect humans because they could preferentially bind to the human-type α−2,6 sialic acid receptors [15]. Of note, the pdm/09 internal gene cassette gradually evolved and may act as the TR internal gene (TRIG) cassette to accommodate the surface genes from different origins. More and more novel genotypes with pdm09 internal gene cassettes were found in pigs [6, 16, 17], which increased the risk of pdm09-reassorted viruses with antigenic diversity infecting humans.

Pigs in China are not vaccinated against influenza viruses. The high development of pig husbandry by increasing the number of large-scale pig farms as well as small holdings for raising animals have provided greater opportunities for the transmission and reassortment of IAVs. During 2013–2015, co-circulation of multiple swIAVs was detected in pigs in Guangxi, including “pure” EA H1N1 and pdm09/H1N1 as well as pdm09-reassorted EA H1N1 and HL H3N2 viruses [10]. Here, we continued the virological surveillance by collecting nasal swabs and lung tissue samples between March 2018 and July 2023. We used the sequences available in the GenBank and Global Initiative on Sharing All Influenza Data (GISAID) databases from Guangxi and analysed the genetic characterization of the pdm/09-reassorted IAVs circulating in the same region after almost ten years.

Materials and methods

Samples collection and virus isolation

A total of 2540 samples, including 2353 nasal swabs and 187 lung tissues were collected from 18 pig farms located in 12 prefectures (Baise, Yulin, Guigang, Guilin, Nanning, Wuzhou, Qinzhou, Liuzhou, Laibin, Beihai, Chongzuo and Fangchenggang) of Guangxi province from March 2018 to July 2023 (Table 1). Samples were collected in sterile phosphate buffer saline (PBS) containing antibiotics and stored at low temperatures until they were shipped to the laboratory. The IAV-positive samples were inoculated into cultures of MDCK cells for isolation using described methods [18].

Table 1. Summary samples collected from pig farms in Guangxi, China, from March 2018 to July 2023.

Prefecture (abbreviation)	No. of samples	No. (%) of positive samples	Isolation
Nanning (NN)	2006	122 (6.07%)	7H1N1, 7H3N2, 1H1N2
Chongzuo (CZ)	265	26 (9.81%)	-
Laibin (LB)	73	12(16.4%)	1 H1N1
Liuzhou (LZ)	46	13 (28.26%)	-
Beihai (BH)	32	5 (15.63%)	-
Wuzhou (WZ)	24	0	-
Guilin (GL)	21	0	-
Guigang (GG)	12	4 (33.33%)	1 H3N2
Yulin (YL)	24	5 (20.83%)	1 H1N1, 1H1N2
Qinzhou (QZ)	11	2 (18.18%)	-
Baise (BS)	12	3 (25%)	-
Fangchenggang (FCG)	14	0(0%)	-
Total	2540	192 (7.56%)	9H1N1 8H3N2 2H1N2

- indicates no isolated strains.

Full-length genome sequencing and assembly

Total viral RNA extract from virus isolates was used to amplify the IAV genomes by reverse transcription PCR (RT–PCR), as previously described [10]. Partial IAV genome sequences were obtained using next-generation sequencing (NGS), as follows: total viral RNA was extracted and used to amplify the whole IAV genomes by using specific primers (Opti1-F1: 5’ GTTACGCGCC AGCAAAAGCAGG, Opti1-F2: 5’GTTACGCGCCAGCGAAAGCAG and Opti1-R1: 5’GTTACGC GCCAGTAGAAACAAGG), by following the instructions of the TaKaRa PrimeScript™ One Step RT–PCR Kit (Dye Plus). The PCR products were then purified using the Agencourt AMPure XP (Beckman Coulter, USA). DNA libraries were constructed using the ALFA-SEQ DNA Library Prep Kit, and whole genome sequencing was performed using the Nova PE150 mode of the Illumina Novaseq 6000 platform. Soapnuke (v1.5.6) [19] was used to filter the raw data to obtain high-quality clean reads, and then SOAPaligner (v2.21) [20] was used to compare the clean reads with the host genome, and sequences whose lengths were less than 80% of the reads were removed. Then, sequence assembly was carried out by using a de novo assembly approach, and the nucleotide sequences were mapped to the reference influenza sequences retrieved from NCBI. 19 full-genome sequences of swIAVs were uploaded in GISAID with the following accession numbers: EPI_ISL_18950379, EPI_ISL_18968262∼3, EPI_ISL_18968286, EPI_ISL_18968291∼3,EPI_ISL 18968298 and EPI_ISL_18968301 to EPI_ISL_18968311.

Phylogenetic analysis and genotype assignment

A dataset included sequences of human, swine and avian IAVs sequences were downloaded from both the Influenza Research Database (IRD) (https://legacy.fludb.org/) and the GISAID EpiFlu database (https://platform.gisaid.org/epi3/cfrontend) and they are listed in Appendix 1 of the Supporting Information. We aligned gene segments individually by using MAFFT v7.27 (https://mafft.cbrc.jp/alignment/ software/) and further edited the data using the MEGA version [21, 22]. Then we constructed phylogenetic trees by using a BEAST 1.8 programme applying a Bayesian Markov Chain Monte Carlo (BMCMC) algorithm. The genotypes of swIAVs were determined by assigning each segment to specific lineages using the maximum-likelihood phylogenies and the genotypes were characterized based on the clade distribution of their internal segments [16]. A strict clock model with coalescent constant population size and a GTR model of nucleotide substitution with gamma-distributed rate variation among the sites were also used. The BMCMC chain lengths were 50 million generations, with sampling every 1,000 generations. The convergence of the parameters was evaluated with Tracer 1.7 (ESS > 200). The tree iteration was discharged with 10% of the chains as a burn-in pattern by using a Tree Annotator (the BEAST package comes with this software). The resulting maximum branching confidence (MCC) tree was then drawn in FigTree version 1.4.4 (http://tree.bio.ed.ac.uk /software/figtree).

Amino acid Analysis

Genetic analysis of swIAVs was performed by using the MegAlign programme (DNASTAR) to compare the nucleotide and amino acid similarities and phenotype-associated amino acid changes. These included the receptor-binding sites, antiviral drug resistance, virulence determinants and specific host preferences among previous and recent swine isolates from Guangxi, as well as other lineages of human origin.

Antigenic analysis

In order to compare the antigenic properties between swIAVs of different lineages and subtypes, a chicken antiserum against seven swIAVs, including the [A/Swine/Guangxi/BB2/2013 (H1N1)] (Sw/GX/BB2/13), [A/Swine/Guangxi/JG24/2019 (H1N1)] (Sw/GX/JG24/19), [A/Swine/Guangxi/JGL1/2020 (H1N1)] (Sw/GX/JGL1/20), [A/Swine/Guangxi/CZ7/2014 (H1N1)] (Sw/GX/CZ7/14) and [A/Swine/Guangxi/JGS17/2019 (H1N2)] (Sw/GX/JGS17/19) were produced in our laboratory. Haemagglutinin inhibition (HI) tests were carried out using this chicken antiserum, by following the WHO guidelines. Analysis of its antigenic properties was performed using the antigenic cartography method as described previously [23].

Results

Virus isolation and genotype assignment

From March 2018 to July 2023, we collected 2540 samples from 18 disease-associated pig farms of 12 prefectures in Guangxi (Figure 1(A)). Of the 2540 samples, 192 (7.56%) were found to test positive for IAVs by RT–PCR (Figure 1(B) and Table 1). 17 of the 18 pig farms had outbreaks of respiratory disease with fever, cough, nasal sneeze and dyspnoea. Four of them had co-infections with other swine pathogens (Supplementary Table S1). We isolated 19 swIAVs, including 9 H1N1 viruses, 8 H3N2 viruses and 2 H1N2 viruses (Table 1). We performed a genome-wide genetic evolutionary analysis of 140 SIVs from Guangxi and found there were multiple lineages and subtypes of swIAVs co-circulating in Guangxi and these were classified into 25 different genotypes (G1–G25) (Figures 2(A,B)). The most prevalent lineage identified in the IAV isolates during the 10 years of this study (2013–2023) was EA H1N1 (G4, 86.67%). This contained HA and NA of the EA H1N1 lineage, internal gene segments (PB2, PB1, PA, NP and M) of pdm09/H1N1, and the NS of CS lineages (Figures 2(B,C)). In contrast, HL H3N2 swIAVs were detected for the first time in 2010 and these have been circulating in pig populations until now [13]. Importantly, there was a high genetic diversity with the changes of surface genes from the recent HL and/or A/Swine/Texas/4199-2/1998 (H3N2) (TX98-like) and A/Swine/Ontario/33853/2005 (H3N2)-like (ON05-like) TR viruses. In particular, a Sw/GX/JG17/19 isolate classified into genotype 18 was a novel quadruple reassortant H1N2 virus with HA, PB2, PB1, PA and NP derived from the pdm09/H1N1, NA from TX98-like and M from G1-like lineages of the H9N2 avian influenza virus as well as the NS from the CS lineage (Figure 2(C)). Obviously, the pdm/09-internal gene cassette became predominant and highly compatible with other swIAVs (Figure 2(D)), suggesting this type of cassette posed a potential public health risk, particularly reassorting with external glycoproteins of distinct antigenicities.

Figure 1. swIAVs identified in 12 prefectures of Guangxi in 2018–2023. (A) The location of the Guangxi Autonomous Region in China. (B) The 12 prefectures in Guangxi from which samples were collected from pigs for this study are shown in the colours used in panel B. The two prefectures shown in grey were not sampled. The prefecture’s two-letter acronyms are identical to those listed in Table 1. Each coloured circle represents a single clinical sample obtained from pigs for this study. The swIAV-positive samples in all locations are coloured in red.

Figure 2. Multiple genotypes and lineages of swIAVs in Guangxi, China. (A) Multiple lineages from swine, human and avian lineages of swIAVs which co-circulated in Guangxi (GX). (B) Genotypes of the different swine influenza A virus (swIAV) identified in Guangxi, including the previously described genotypes and obtained in this study. The eight bars represent the eight gene segments and the colours of the bars indicate the lineage of origin of the gene segments, corresponding to Figure 2A. (C) Genotypes of EA H1N1, HL H3N2 and pdm09-like H1N2 swIAVs in this study. (D) Contributions (%) of various genetic lineages to gene segments in the 158 reassortant swIAV strains in Guangxi during 2011 and through 2023.

Natural reassortment of a novel swIAV isolate carrying the NA gene from TX98-like or ON05-like TR virus and the M gene from avian H9N2 virus

On November 18, 2019, a novel quadruple reassortant [A/Swine/Guangxi/JGS17/2019(H1N2)] was detected in breeding pigs with severe respiratory disease, which showed symptoms of fever, dyspnoea and loss of appetite (Supplementary Table S1). Analysis of the genotype characterization indicated that their HA gene belonged to pdm09/H1N1 lineage (Figure 4), but it was distinct from the early pdm09/H1N1-like strain (Sw/GX/DX24/13), only sharing 95.2% nt identity. Interestingly, it was found for the first time that the NA gene segment shared high identities with the TX98-like virus circulating in pigs in 1998, which resulted in several outbreaks of severe respiratory disease in the USA [24] (Supplementary Figure S1). Notably, phylogenetic analysis based on the M gene showed that this novel H1N2 swIAV isolate carried a M gene segment from the avian G1-like H9N2 lineage, which was genetically associated with the human H7N9 strain [A/Jiangsu/05155/2016(H7N9)] (Figure 3), sharing 98.9% nt identity. This suggested that the introduction of the avian M gene into pigs could have originated from H7N9-infected humans. Importantly, the reverse zoonotic infection from humans to pigs will enrich the endemic swIAVs gene pool and will likely generate novel reassortant viruses that may be continually transmitted between these two species.

Figure 3. Phylogeny and divergence time of M genes of swIAVs isolates in Guangxi. The phylogenetic tree of the M gene was generated by using the Bayesian MCMC framework, using the GTR substitution model with gamma-distribution among the site rate heterogeneity and a “strict molecular clock” model. The branches are coloured according to their host origin.

Genetic diversity of swIAVs in Guangxi during 2013–2023

To determine the overall genetic characterization of the H1N1, H3N2 and H1N2 swIAVs in Guangxi during 2013–2023, we downloaded 140 HA (H1 and H3) and 140 NA (N1 and N2) swIAV sequences detected in Guangxi from FLU database (Supplementary Table S2). These were combined with the 19 swIAV sequences obtained in our study, and we conducted a molecular clock phylogenic analysis and genotype characterization.

• (i). H1 tree. Phylogenetic analyses of the H1 haemagglutinin (HA) gene revealed that these viruses clustered into four lineages in Guangxi, including the CS, TR H3N2, pdm09/H1N1, and EA lineages (Figure 4). The EA H1 HA lineage was predominant in pigs. Most of the contemporary EA H1 swIAVs in the study formed a subclade, which was genetically distinct from the pre-2015 swine isolates. In particular, most of the swIAV isolates found between 2021 and 2022 were grouped with the human EA H1N1 strains. It should be noted that the Sw/GX/JGL1/2020 (H1N1) virus shared a high identity (98.2%) with the human strain (A/Yunnan-Mengzi/1462/2020), indicating the possibility that the introduction of the reassortant EA H1N1 viruses from pigs into humans had already been established. The pdm09-like H1N1 had been detected since 2011 and was sustained in pigs for > 3 years. In this study, it reassorted with other prevailing viruses of the TX98-like or ON05-like TR lineage and generated a novel quadruple or triple H1N2 virus (Sw/GX/JGS17/19), which harboured the M gene segment from avian H9N2 virus. Interestingly, this novel virus displayed longer branch lengths than in the pdm09/H1N1 lineage of the HA trees, reflecting the pdm09-origin HA gene had diverged, being genetically different and distinct from the pre-2015 strains.

Figure 4. Phylogeny and divergence time of H1 HA genes. The phylogenetic tree of the H1 HA gene was generated by using the Bayesian MCMC framework, using the GTR substitution model with gamma-distributed among site rate heterogeneity and a “strict molecular clock” model. Coloured boxes show the lineage classification of the HA gene segments of IAVs. The isolates in this study are labelled with solid red circles. Blue and green solid circles indicate early isolated strains of Guangxi and EA H1N1 strains isolated from humans, respectively.

Based on the H1 lineage classification, there were 19 genotypes of G1-G19 found in H1 viruses in Guangxi (Figure 2(B); Figure 6). There were also “pure” viruses from the same lineage (e.g. EA, CS, pdm09/H1N1 lineage) detected in pigs in the pre-2013 survey regions and these were designated as G1, G13 and G20, respectively. Subsequently, they were replaced with the pdm09-EA H1N1 reassortant viruses, generating G3-G10 viruses between 2013 and 2019. Among these pdm09-EA H1N1 reassortant viruses, the G4 virus was the most prevalent in pigs in Guangxi from 2013 to 2022. It was found that they had spilled over from pigs to humans, indicating the potential risk to human health. In addition, the N2 NA genes from early H3N2 TR viruses were detected frequently in the H1N2 subtype swIAVs and this generated seven genotypes (G11 and G14-G19), which included the introduction of the M gene derived from the G1-like lineage of avian H9N2 virus (G19).

• (ii). H3 tree. The “pure” HL H3 subtype of swIAVs was detected in Guangxi before the 2000s, while a novel reassortant HL H3N2 virus carrying pdm09-like internal genes was only detected in Guangxi in November 2011 [13]. Based on the time scale of the MCC tree, the swine HL H3N2 lineage appeared to have diverged from the human seasonal H3N2 viruses and was further divided into two divergent sub-lineages, including recent HL and early TR TX98-like (Figure 5). Phylogenetic analysis indicated the HA gene of a Sw/GX/JGX3/20 (H3N2) isolate was genetically close to the A/swine/Texas/4199-2/1998 (H3N2) virus, sharing 99.1% nt identity. The remaining 7 H3N2 viruses were grouped with an early HL H3N2 virus from Vietnam (Sw/BinhDuong/03_06/10, BD-like). Of note, five of the swIAV isolates formed a subclade with human strain [A/Guangxi-Gangbei/190/2019 (H3N2)], which was isolated from an unvaccinated two-year-old girl in Guigang City, Guangxi, in January 21, 2019. The homology of the HA gene of the Sw/GX/JG13/2019 isolate shared a homology of 98% at the nucleotide level, indicating that these reassortant H3N2 viruses had the zoonotic potential to infect humans.

Figure 5. Phylogeny and divergence time of the H3 HA genes. The phylogenetic tree of the H3 HA gene was generated by the Bayesian MCMC framework, using the GTR substitution model with gamma-distributed among site rate heterogeneity and a “strict molecular clock” model. Coloured boxes show the lineage classification of HA gene segment of IAVs. The isolates in this study are labelled with solid red circles. Blue and green solid circles indicate early isolated strains of Guangxi and Ontario-like H3N2 strains isolated from humans, respectively.

Based on the H3 lineage classification, there were six genotypes of G20-G25 found in the H3 viruses from Guangxi (Figure 2(B) and Figure 6). G21 viruses with HA and NA genes from the HL lineage and the remaining gene fragments from the pdm09/H1N1 lineage were dominant in pigs in Guangxi from 2010 to 2011. The NS gene of pdm09-HL swIAVs was gradually replaced by the CS lineage and formed the G23 virus, which became endemic in Guangxi from 2012 onwards. Notably, the H3 HA gene and N2 NA derived from the TX98-like or ON05-like H3N2 TR lineage were frequently introduced into pdm09-HL H3N2 reassorted viruses and replaced the surface genes, which produced G24 and G25 viruses.

Figure 6. Genesis of the reassortant genotypes of SIVs that became established in pigs. Viral particles are represented by coloured oval shapes containing the lineage of origin (the colours of the lineage origin of segments are shown in Figure 2(A)). The coloured long bars indicate the gene segments, from top to bottom, PB2, PB1, PA, HA, NP, NA, M and NS. The segments in the progeny viruses are coloured according to the source viruses (top). The arrows indicate the flow from the source virus of a gene segment (numbered at the arrow tails) to the receiving reassortant virus (arrowheads). The reassortant viruses in this study are shaded in grey. The timeline on the right indicates the year when the novel reassortant viruses were found.

Antigenic analysis among early and recent H1 swine isolates

To assess the antigenic properties of these reassortant H1N1 viruses, ten representative H1 viruses, including two early EA H1N1 viruses, seven contemporary EA H1N1 isolates and one H1N2 isolate were selected to carry out the HI assay using chicken antiserum. In these tests, the EA H1N1 Sw/JGL1/20 was able to react well with early and recent EA H1N1 viruses, and pdm09-like H1N2 virus, whereas there was four-to sixteen-lower reactivity between early and recent isolates, indicating some antigenic heterogeneity among the reassortant H1 swIAV isolates (Supplementary Table S3). Further analysis showed several amino acid differences in the HA antigenic sites including 144 (H3 numbering) in Ca2 and 75 in Cb, and other consistent mutations at the positions 51, 72, 181, 303, 63, 497, 514, 527 among previous and other recent isolates were observed (Supplementary Table S4).

Differences in specific host and virulence markers

We analysed and compared these swIAVs isolated in pre-2015 isolates and those from recent years. For H1 HAs, the HA receptor binding sites in all EA H1N1 from G1, G4, G8, G13 and G19 were still found to be highly conserved. They contained 190D, 223 V, 225E, 226Q and 228G, while the pdm09-like virus (Sw/GX/JGS17/19) contained the same mutation as the pre-2015 viruses, except for the mutation of 190A, which was consistent with the human strain (A/Guangxi-Gangbei/SWL 11063/201) (Supplementary Tables S4). It was determined that the amino acids of 226Q/228G in H1 HA had the potential to bind the human-type α−2,6 sialic acid receptors.

As for H3 HAs, there were two mutations of D190N and V223I in the HA protein of four HL H3N2 isolates, but they persistently kept isoleucine (I) and serine (S) at positions 226 and 228, which were determined to have specificity for the human-type α−2,6 sialic acid receptors [25, 26]. In addition, it was found that the two RBD sites at position 135 and 193 and the two antigenic sites (e.g. D144N and L160I) showed antigenic drift in the recent swIAV isolates, suggesting HL H3N2 swIAVs underwent some selected pressure in swine herds over the ten years (Supplementary Tables S5). The pdm09-like internal gene cassette has been predominant in these reassortant viruses in recent years. The majority of swIAVs isolates kept 271A, 588I, 590S, 591R, 627E and 701D, which enhanced the pathogenicity of the pdm09/H1N1 virus and its adaptation in mammals. However, there were two swIAVs isolated in 2019 and 2020, which had a threonine (T) at position 588 and a glycine (G) at position 590 [27]. In addition, EA H1N1 viruses possessing the NA-274H and – 294N substitutions were also observed, indicating their susceptibility to the antiviral drug, oseltamivir. All the H1N1 and H3N2 isolates contained the adamantane-resistant mutation, S31N, in the M2 protein (Supplementary Table S6).

Discussion

There is a complicated and diverse ecosystem of influenza viruses in southern China, where there is a common co-circulation of H1N1, H1N2 and H3N2 subtypes and multiple lineages swIAVs in the pig herds associated with this region [6, 28, 29]. Here, systematic monitoring and assessment of the potential risk of influenza virus’s emergence in pigs in Guangxi during 2018–2023 revealed multiple lineages of swIAVs co-circulated in the pig populations, including the H1N1, H3N2 and H1N2 subtypes. Among the multiple reassortment viruses, G4 EA H1N1 was still found to be dominant in Guangxi and this has persistent for 10 years (Figure 6). Strikingly, the genetic diversity of H3 swIAVs in Guangxi increased since 2020. We found that the NA gene of the HL H3N2 viruses was gradually replaced by the ON05-like virus of the TR H3N2 lineage from Canada. Importantly, a human isolate, A/Guangxi-Gangbei/190/2019, was clustered with these novel H3N2 swine viruses. It is possible that the persistent novel H3N2 reassortants may be in the process of replacing the recent HL H3N2 in swIAVs. We also isolated for the first time, a novel quadruple reassortment between pdm09-like H1N1 and TR H3N2, which contained an avian H9N2 M gene segment that originated from the introduction of human H7N9 viruses through reverse zoonosis into pigs (Figure 3). Recently, Sun et al. (2022) detected a swIAV with a natural reassortment between the avian H9N2 and EA H1N1 viruses, and these included the PB1 in H10N8 and M in H7N9 [30]. In fact, previous reports indicated pigs could be infected by the H7N9 virus without any clinical symptoms [31]. Jones et al. (2013) assessed the replication ability of human H7N9 viruses in swine tissue explants and found that they could replicate efficiently in tracheas and bronchi [32]. Xu et al. (2014) also tested the mammalian host adaptation of the human H7N9 virus in pigs through passaging and indicated that the H7N9 virus could infect pigs although there was poor replication ability after three passages in pig lungs [33]. These results suggested the possible reassortment between the endemic swIAV and H7N9 viruses. However, the quadruple H1N2 swIAV carrying avian H9N2-M gene did not replicate efficiently in the MDCK cell, suggesting that the low genetic compatibility between avian H9N2-M gene and other genes may limit the replication in pigs. Notably, except for the introduction of the avian H9N2-M gene segment, repeated introductions of the surface genes from TR H3N2 TX98-like and ON05-like viruses also resulted in expanding the swIAV gene pool in Guangxi. This has led to the co-circulation of cross-lineage reassortant viruses in the field and whether these will facilitate the onward transmission from pigs to humans needs to be closely monitored.

All the pdm09/H1N1 viruses and their surface genes could not persist in pigs. In particular, the pdm09-origin surface genes were not sufficiently competitive with the enzootic swIAVs but the pdm09-origin internal genes appeared to have long-term persistence [6]. However, the combination of pdm09-originated PB2, PB1, PA and NP and CS-originated NS genes (pdm09/CS) had an obvious competitive advantage in pigs, which could be highly compatible with different surface genes of the enzootic swIAVs. The HA gene of EA H1N1 viruses introduced into China belonged to the 1C.2.3 clade [34], which was predominant in China and formed multiple genotypes through reassortment with the endemic swIAVs circulating in pigs [11, 12, 35]. They had a typical feature with their vRNP being derived from the pdm09/H1N1 virus and the NS gene from the CS virus (Supplementary Figure S2). Moreover, the differences observed in the M gene-originated viruses resulted in the emergence of two different genotypes. Meng et al. (2023) indicated that the two genotypes were predominant in pigs [11], but our surveillance in recent years only found the G4 EA H1N1 viruses in pig farms with severe respiratory disease and high morbidity rates. In accordance with the swIAV outbreak, many of the workers in pig farms also showed flu-like symptoms. The available evidence indicated that 10.4% (35/338) of the workers were positive for G4 EA H1N1 virus [9]. Of note, studies have shown that the EA H1N1 viruses were preferentially bound to human-like SAα 2,6 Gal receptors while they showed enhanced pathogenicity in the mouse and ferret models [9, 36, 37]. Clinically, there were two cases where patients were infected by G4 EA H1N1, resulting in severe clinical syndrome, and even death [38, 39]. Our studies showed that the G4 EA H1N1 viruses persisted for almost ten years and greatly affected the prevalence of currently enzootic swIAVs in Guangxi, and even resulted in cases of human infection. Whether the continued antigenic drift and expansion of the genotypic diversity through reassortment with enzootic H3N2 or H1N2 swIAVs in the same region will change the evolution patterns and ecosystem of swIAVs in Guangxi still remains to be seen. However, this data emphasizes the importance of long-term surveillance for EA H1N1 viruses at the pig-human interface.

The HL H3N2 swIAVs were detected in Guangxi in 2010 [13] and they spread to Guangdong and Hong Kong in mid-2011 [6]. Most of the introduced HA/NA segments of human seasonal influenza viruses have become established in pigs with apparently onward transmission. Our studies revealed that although HL H3N2 persisted in pigs for ten years, the repeated introduction of the N2 from the TR H3N2 lineage changed the genotype diversity of swIAVs. The N2 NA gene segments of these novel H3N2 viruses were associated with the TR H3N2 lineage, an A/Swine/Ontario/33853/2005(H3N2) virus that originated from Canada, which formed the G25 viruses. Unfortunately, the pig-to-human spillover of G25 viruses occurred in an unvaccinated two-year-old girl in 2019, and this showed that these viruses can pose a threat to public health once they are capable of efficient transmission to humans. In addition, this was the first time that the introduction of the H3 HA and N2 NA segments from the early TR H3N2 lineage (TX98-like) into the recent HL H3N2 and pdm09-like H1N1, respectively, was detected. This resulted in the production of four novel genotypes, G11, G19, G24 and G25.

Sequence analysis found that there were 1-amino acid (lysine) insertions between positions 46 and 47 of the NA stalk region, and this was reported in 2012 [6]. Except for the insertion at this position, there was only one other report that a Sw/GX/JGX3/2020 isolate showed a two-aa insertion (lysine-lysine) observed between 74 and 75 of the NA stalk regions (Supplementary, Figure S3). A short – and long-stalk NA protein may influence the biological functions of IAVs. Bi et al. (2016) also reported that a 5-amino-acid deletion of the N9 stalk in the H7N9 virus could enhance its virulence in mice [40]. The additional glycosylation of the HA protein and a short-stalk NA in the H5N1 virus also appeared to contribute to its virulence in mice [41]. Lin et al. (2016) also demonstrated that the long stalk of NA (with a 2-aa insertion) in the H3N2 canine influenza virus exhibited a higher viral replication rate and may also have facilitated a more efficient transmission in dog populations [42]. The more biological significance of the sequence alterations in these viruses needs to be further investigated.

Guangxi province is located in the south of China where there were more than 30 million pigs in 2022 and this region produces approximately 2.2 million tons of pork. International trade and movement of animals bring further unexpected introduction of the earlier existing TR H3N2 swIAVs into the locally circulating swIAVs. It is worth noting that the TX98-like and ON05-like viruses were introduced into pig populations via imported animals. The NA gene from the human seasonal H3N2 or EA H1N1 viruses was replaced by the early TR H3N2 lineage (ON05-like) and these formed new HL H3N2 viruses, resulting in the generation of four novel genotypes (G11, G19, G24, G25). Importantly, the introduction of the H9N2 G1-like M gene into the pdm09/H1N1-like virus (G19) generated a novel quadruple reassortant H1N2 virus. These findings further emphasize the importance of active surveillance for the local and imported swine herds in order to uncover the emergence of novel swIAV strains that could pose potential threats to human health.

Acknowledgements

We thank Dr. Dev Sooranna of Imperial College and YMUN for the English language editing of the manuscript. Chongqiang Huang, Liangzheng Yu, and Yi Xu contributed equally to this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability

Accessions with relevant swine influenza viruses obtained from GISAID and IRD are included in Supplementary Tables S2.

Additional information

Funding

This work was supported by the NIAID supported Centers of Excellence for Influenza Research and Response (CEIRR, contract no. 75N93021C00014), Guangxi Natural Science Foundation (grant number 2023GXNSFDA026042) and the National Natural Science Foundation of China (32160825).